Otimização do controle do oxigênio dissolvido em biorreator airlift pressurizado: aplicação em cultivos de Escherichia coli recombinante

Abstract

A wide variety of industrial and therapeutic proteins is synthetized by genetically modified Escherichia coli, which is ease to cultivate and manipulate and also well characterized. However, there are only few studies regarding E. coli cultivation in the pneumatic airlift bioreactor. This reactor presents some advantages over the stirred tank, such as simpler construction, lower risk of contamination, and efficient gas-liquid dispersion with reduced power consumption. However, the lower oxygen transfer capacity in the bench-scale airlift bioreactor, in comparison to the stirred tank, justifies the use of temperature, pressure, and gas and oxygen flow rates as manipulated variables for the dissolved oxygen (DO) control. In this context, this thesis aims: (i) to develop a mathematical model that describes the production of the Pneumococcal Surface Protein A (PspA) by recombinant E. coli in a pressurized airlift reactor, taking into account oxygen transfer and uptake in the process; (ii) to perform economic optimization of the DO control; (iii) to develop an advanced DO controller integrated to state estimators. Data of E. coli cultivation in conventional and airlift reactors were used to identify and validate the models. The dynamic optimization was performed using a gradient method based on the Pontryagin's minimum principle. The developed mathematical models were able to describe the process under varying conditions of temperature and pressure. The dynamic optimization of the DO control resulted in a simple and sequential way of handling the inputs: manipulation of air flow rate, followed by system pressurization, and then air enrichment with pure oxygen. The optimum process temperature was 27 °C. A model predictive DO control was then proposed, associated to two state estimators, the extended Kalman filter (EKF) and the moving horizon estimator (MHE). They were both able to estimate satisfactorily four state variables (cell, substrate, PspA, and DO concentrations) based only on DO measurements. The control system proved to be robust in process simulations. The conclusions that arise from this thesis contribute to the area of development of non-conventional reactors and DO controllers, especially for pneumatic bioreactors, which are widely used in aerobic bioprocesses.